초록 ▼

In a process for detecting target patterns in a texture with which a system state can be represented as a distribution of points in an n-dimensional space, anisotropic scaling factors a.sub.ji are determined which are characteristic of the relationship between the projection of the point density ont

In a process for detecting target patterns in a texture with which a system state can be represented as a distribution of points in an n-dimensional space, anisotropic scaling factors a.sub.ji are determined which are characteristic of the relationship between the projection of the point density onto coordinate j and the distance to examined point P.sub.i. A target is detected if the distribution of scaling factors deviates from predetermined comparison distributions of systems without target patterns.

대표청구항 ▼

[ What is claimed is:] [1.] A process for detecting target patterns in a texture with which a system state can be represented as a distribution of points (P.sub.1) in an n-dimensional space, the process comprising the steps of:a) for each point (p.sub.i) having coordinates (x.sub.1,i . . . x.sub.n,i

[ What is claimed is:] [1.] A process for detecting target patterns in a texture with which a system state can be represented as a distribution of points (P.sub.1) in an n-dimensional space, the process comprising the steps of:a) for each point (p.sub.i) having coordinates (x.sub.1,i . . . x.sub.n,i) determining at least one point number M.sub.ji (d.sub.2), which comprises all points at the time, with coordinate j in an interval [x.sub.ji +d.sub.2, x.sub.ji -d.sub.2 ], and determining at least one point number M.sub.ji (d.sub.1) with coordinate j in an interval [x.sub.ji +d.sub.1, x.sub.ji -d.sub.1 ], wherein d=d.sub.2 -d.sub.1 is a width of an interval about coordinate (x.sub.j,i) of point (p.sub.i)b) for each point (p.sub.i), determining at least one scaling factor a.sub.ji such that EQU a.sub.ji =[log M.sub.ji (d.sub.2)-log M.sub.ji (d.sub.1)]/[log d.sub.2 -log d.sub.1 ],c) determining a plurality of windows of size k which each comprise a subregion of the distribution of points p.sub.i,d) for said plurality of windows, determining a plurality of probability densities P.sub.1 (a.sub.ji) for each said scaling factor a.sub.ji accordingly,e) determining at least one of an average value and a standard deviation of each said probability density P.sub.1 (a.sub.ji), andf) identifying an occurrence of a target pattern if probability density P.sub.2 of at least one of said average value and said standard deviation deviates from a reference texture comparison value.